AI hardware,held in stock.
H200, RTX PRO 6000 and L40S GPUs, Supermicro servers, DDR5 memory and 400G/800G optics. Sourced through manufacturer channels, held in Dubai and Hong Kong, quoted within one business day.
Categories
From a single GPU to a full rack.
01GPUs & acceleratorsH200, L40S, RTX PRO 6000 — server and workstation
02AI serversSupermicro, Dell, HPE GPU systems, 4U–8U
03MemoryDDR5 RDIMM 5600–6400, 64 GB modules
04Networking400G / 800G optics, InfiniBand and Ethernet
05AI workstationsDGX Spark, RTX PRO desktops
06ComponentsPower supplies, NVMe drives, cooling and spares
In stock now
On the shelf in Dubai and Hong Kong.
Read live from warehouse inventory. Stock is shown as a quantity tier; exact quantities and pricing come with the quote.
How stock moves
One chain of custody, from allocation to your dock.
01
Source
Allocation through NVIDIA, Supermicro, Dell and HPE channels and their authorised distributors. Every lot arrives with origin paperwork, not a broker’s word.
02
Verify
Units are received at the bench, inspected and logged by serial number before they are offered. Configured systems are assembled and tested before dispatch.
03
Ship
Dispatched from bonded stock in Dubai or Hong Kong, with commercial invoice, packing list and certificate of origin ready for your broker.
Ordering
How ordering works
No accounts and no checkout. A quote is a document you can take to finance.
- 01
Send a list
Model names, part numbers or a short platform brief. A quote list built on this site works the same way.
- 02
Get a firm quote
Unit price, lead time, warranty terms and delivery basis, within one business day.
- 03
Confirm and allocate
Units are reserved against your order and their serial numbers are recorded before dispatch.
- 04
Ship with documents
From Dubai or Hong Kong, with commercial invoice, packing list and certificate of origin.
Hubs
Two hubs, one inventory.
Stock is held in Dubai and Hong Kong and listed as one catalog. An order ships from whichever hub holds the units, with export documents prepared at origin.
- Headquarters
- Dubai, UAE
- Second hub
- Hong Kong
- Documents
- Invoice · Packing list · CoO
- Stock view
- One catalog, updated daily
DXB · Dubai HQ25.20°N 55.27°EStock · Sales · Export documents
HKG · Hong Kong22.32°N 114.17°EStock · Asia-Pacific dispatch
DXB ↔ HKG50°E60°E70°E80°E90°E100°E110°E120°EManufacturers
Channels we buy through.
- NVIDIAH200 · RTX PRO 6000 · L40S · DGX Spark
- AMDInstinct accelerators · EPYC
- SupermicroGPU servers, 4U–8U
- DellPowerEdge XE GPU servers
- HPEProLiant and Cray GPU systems
- SamsungDDR5 RDIMM
- MicronDDR5 RDIMM
- KingstonServer Premier DDR5
- MellanoxInfiniBand and Ethernet
- BroadcomEthernet switching and NICs
- IntelXeon · Gaudi · Ethernet adapters
- Source Photonics400G / 800G optics
Insights
Notes from the bench.
- Choosing GPUs for LLM inference vs trainingInference is a memory problem, training is a bandwidth-and-interconnect problem. How to size VRAM, read precision figures and decide between PCIe cards and NVLink systems.
- H200 NVL vs RTX PRO 6000 vs L40S: where each winsThree PCIe cards, three price bands. A side-by-side of memory, bandwidth, power and cooling, and the workloads where each is the right answer rather than the expensive one.
- Memory sizing for AI servers: DDR5 RDIMM rules of thumbHow many channels to populate, how much system RAM per GB of VRAM, and which speed bins and ranks to specify so the memory keeps up with the GPUs.
Next step
Send the list. Get a firm quote within one business day.
Part numbers, quantities and a delivery country are enough to start. Pricing, lead time and documents come back in one reply.